Press "Enter" to skip to content

Day: June 25, 2020

Using INLA for Spatial Regression in R

Lionel Hertzog continues a series on spatial regression:

INLA is a package that allows to fit a broad range of model, it uses Laplace approximation to fit Bayesian models much, much faster than algorithms such as MCMC. INLA allows for fitting geostatistical models via stochastic partial differential equation (SPDE), a good place for more background informations on this are these two gitbooks: spde-gitbook and inla-gitbook.

This is not the gentlest introduction, so if you’re new to the concept go back and read part 1.

Comments closed

Thoughts on Snowflake Database Provisioning

David Stelfox takes us through some thoughts on provisioning instances of Snowflake:

For this example, I’ve chosen an open dataset of 2017 taxi rides in New York City. There are a few options for interacting with Snowflake: a dialog box approach in the web-based GUI, using SQL statements in the Worksheets tab in the GUI or a CLI called SnowSQL. For this example, I used SQL statements as I find them easier to follow what’s happening. Once you have set up your account (or trial) and logged in, you need to create your first database.

Click through for some how-to as well as thoughts about cost and performance.

Comments closed

Accelerated Database Recovery and tempdb Usage

Jason Hall takes a look at how much tempdb space Accelerated Database Recovery might use:

You might have heard me talk about tempdb parasites in the past, or maybe you’ve read my blog post on the same topic, “Be Mindful of SQL Server Tempdb Use (aka Tempdb Parasites!).” I know that at least one person did, because they recently asked a great question.

In that blog post, I reviewed how triggers use the version store in tempdb for access to the “special” trigger tables we can use from within the trigger code to access the previous and new versions of data being modified. One person on Twitter was wondering if that is still the case for triggers on databases using Accelerated Database Recovery (ADR) in SQL Server 2019. I really wasn’t sure, so I decided to find out.

Click through for the answer.

Comments closed

Distributed Transactions Across Multiple Log Files

Eric Cobb notices something strange:

When the new log files were added, SQL Server immediately filled them to match the fullness percentage of the original log files. As I manually increased the size of the log file, SQL Server again shifted the transactions around to keep both log files at the exact same fullness percentage. So, if the first log was 95% full, the second log was 95% full, regardless of the actual log file sizes. If the first log was 80% full, so was the second. The more I expanded the second log, the more transactions SQL Server would move to it, always keeping the fullness percentage the same on both log files. The larger the second log became, the more space it freed up on the first log, but the 2 logs were always exactly in sync in terms of fullness.

This is some interesting behavior, especially because transaction log files don’t use proportional fill.

Comments closed

Creating a New Container from a SQL Server on Windows Dockerfile

Jamie Wick continues a series on SQL Server and Windows containers:

The docker build command sends the contents of the working directory, along with a dockerfile, to the Docker daemon, as a build context, to create the new image. A dockerfile is a plain text file that contains the name of a (base) image, along with a set of instructions for modifying the image. By default, the dockerfile is assumed to be in the root of the working directory, but a separate location can be specified using the -f parameter in the build command. Additionally, the -t parameter can be used to specify a repository and tag for the new image. Finally, the working directory can be specified using a Path or URL. In the example below, the current directory (.) is being used as the working directory (the docker build command is being run at the root level of the working directory).

Read on for examples.

Comments closed

Shredding XML for Service Broker

Chris Johnson continues a series on Service Broker:

So, what we have here is an element called people. This contains 2 person elements with personID attributes attached to them. These both contain firstName and lastName elements, but personID 124 also has a middleName and two pet elements. These pet elements each hold a name element, but only fluffles has an animalType element.

In this way, we can use XML to hold only the data we actually know. Unless we have specified an XML Schema (outside the scope of this class) there is no validation on what elements or attributes an element should contain.

When Service Broker was launched, XML was the primary markup language people pretended was easily readable but really wasn’t.

Comments closed

Calculating Business Hours with DAX

Matt Allington combines DAX and a calendar table to calculate business hours:

I was helping a client this past week to calculate the total business hours between a start date/time and an end date/time, taking into account the working days, public holidays and non-working weekends, etc.  As is often the case, I thought it would be a great blog article that I could share with my readers.  This is a pretty involved problem with lots of moving parts, so as such I have decided to record a video showing you how I solved the problem, 1 step at a time.

Click through for the video as well as a description and the code.

Comments closed